Understanding Adversarial Transferability in Federated Learning